CN116697933A

CN116697933A - Optical fiber bending angle identification method and device, electronic equipment and storage medium

Info

Publication number: CN116697933A
Application number: CN202310613379.2A
Authority: CN
Inventors: 杨婧雅; 王晓东; 尹祖新; 刘琦; 田洪宁; 赵广; 魏汝翔; 梁芳
Original assignee: China United Network Communications Group Co Ltd
Current assignee: China United Network Communications Group Co Ltd
Priority date: 2023-05-26
Filing date: 2023-05-26
Publication date: 2023-09-05

Abstract

The application discloses a method, a device, electronic equipment and a storage medium for identifying the bending angle of an optical fiber, which relate to the field of optics and can solve the problem that the bending angle of the optical fiber cannot be measured at low cost, in a large range, in multiple directions and with small error at the present stage, and comprise the following steps: preprocessing the optical fiber to be identified, and determining at least one sensitization area; according to the optical experiment platform and the at least one sensitization area, speckle image data of the optical fiber to be identified are obtained; and inputting the speckle image data of the optical fiber to be identified into a trained speckle image identification model, and determining the bending angle identification result of the identified optical fiber. The application is used for identifying the bending angle of the plastic optical fiber.

Description

Optical fiber bending angle identification method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of optics, and in particular, to a method and apparatus for identifying a bending angle of an optical fiber, an electronic device, and a storage medium.

Background

For the recognition of the bending angle of the optical fiber, the existing distributed sensing technology generally uses an optical frequency domain reflectometer and an optical time domain reflectometer, and the resolution precision can reach the centimeter level. However, the architecture of the sensing system using this technology is complex, and the amount of hardware used is large, which results in high cost.

While intensity modulation based bending sensors have the advantages of low cost, simple manufacturing process, etc., they have a significant disadvantage of lacking multiplexing functionality. Specifically, intensity modulation based bending sensors cannot perform multi-point or multi-parameter measurements, which can reduce the flexibility and compactness of the sensing system if multiple optical fibers are used to perform the multi-point or multi-parameter measurements. In addition, most of the bending sensors in the prior art cannot distinguish between directions, or can distinguish between two opposite directions, and only a small portion can distinguish between multiple bending directions, which is insufficient for bending measurement in practical applications.

Disclosure of Invention

The application provides a method and a device for identifying an optical fiber bending angle, electronic equipment and a storage medium, which can solve the problem that the bending angle measurement of the optical fiber cannot be carried out in a low cost, large range, multiple directions and small error at the present stage.

In order to achieve the above purpose, the application adopts the following technical scheme:

in a first aspect, the present application provides a method for identifying a bending angle of an optical fiber, including: preprocessing the optical fiber to be identified, and determining at least one sensitization area; according to the optical experiment platform and the at least one sensitization area, speckle image data of the optical fiber to be identified are obtained; and inputting the speckle image data of the optical fiber to be identified into a trained speckle image identification model, and determining the bending angle identification result of the identified optical fiber.

Based on the technical scheme, the step of polishing and grinding the side surface is designed, so that the processed plastic optical fiber section to be identified becomes a distributed optical fiber sensor for measuring the multi-point multi-plane bending angle, and meanwhile, a speckle identification model is trained by utilizing a convolutional neural network, so that different bending angles and directions of plastic optical fibers can be identified in a classified manner based on a visible light source or a light-emitting diode (LED) lamp. The application does not need expensive optical instruments, and pretreats the plastic optical fiber, so that the optical fiber has extremely high sensitivity to a plurality of planes or directions, and can realize multi-directional bending angle identification.

In one possible implementation, the optical experiment platform includes: the device comprises a bottom panel, an optical clamp, a fixed end, a visible light source, a convex lens, a Charge Coupled Device (CCD) camera and an image acquisition device; the bottom panel is used for fixing the optical clamp, the fixed end, the visible light source, the convex lens, the CCD camera and the image acquisition device; the optical clamp is used for fixing the optical fiber to be identified; the fixed end is positioned on the optical fiber to be identified, and the fixed end and the convex lens are respectively positioned at two sides of the optical clamp; the visible light emitted by the visible light source is emitted into the optical fiber to be identified from the fixed end; the CCD camera is used for capturing emergent light emitted from the convex lens; the image acquisition device is connected with the CCD camera and is used for acquiring speckle image data of the optical fiber to be identified.

In one possible implementation, the speckle image recognition model is constructed from a convolutional neural network, and the trained speckle image recognition model is determined according to the following steps: acquiring a data set, and marking the data set; according to the optimizer and the loss function, adjusting parameters of a speckle image recognition model; the optimizer is used for adjusting the learning rate of the speckle image recognition model; and under the condition that the loss function value meets the preset condition, determining that the speckle image recognition model training is completed.

In one possible implementation, the preprocessing includes: and performing side polishing processing on the optical fiber to be identified, and selecting at least one sensitization area.

In one possible implementation, the at least one sensitization region includes a first sensitization region, a second sensitization region, a third sensitization region, a fourth sensitization region, and each quarter of the fiber to be identified includes a sensitization region; each quarter region of the optical fiber to be identified comprises a sensitization region; the first sensitization area is perpendicular to the plane where the second sensitization area is located, the first sensitization area is parallel to the plane where the third sensitization area is located, and the first sensitization area is perpendicular to the plane where the fourth sensitization area is located.

In one possible implementation, each of the sensitized areas is 20 mm in length and the polishing depth is one quarter of the core diameter of the fiber to be identified.

In one possible implementation, the type of optical fiber to be identified is a dual-core plastic optical fiber; each fiber core of the optical fiber to be identified is polished and ground on the side face, and each fiber core comprises four sensitization areas.

In one possible implementation, before acquiring the speckle image data of the optical fiber to be identified according to the at least one sensitization zone and the optical experiment platform, the method further comprises: coupling the two fiber cores of the optical fiber to be identified; wherein the sensitized areas of the two cores are opposite.

In a second aspect, the present application provides an optical fiber bending angle recognition apparatus, comprising: a processing unit; the processing unit is used for preprocessing the optical fiber to be identified and determining at least one sensitization area; the processing unit is also used for acquiring speckle image data of the optical fiber to be identified according to the optical experiment platform and the at least one sensitization area; the processing unit is also used for inputting the speckle image data of the optical fiber to be identified into the trained speckle image identification model, and determining the bending angle identification result of the identified optical fiber.

In one possible implementation, the optical fiber bending angle identifying device further includes: an acquisition unit; the acquisition unit is used for acquiring the data set and marking the data set; the processing unit is also used for adjusting parameters of the speckle image recognition model according to the optimizer and the loss function; the optimizer is used for adjusting the learning rate of the speckle image recognition model; and the processing unit is also used for determining that the speckle image recognition model training is completed under the condition that the loss function value meets the preset condition.

In one possible implementation, the at least one sensitization region includes a first sensitization region, a second sensitization region, a third sensitization region, a fourth sensitization region, and each quarter of the fiber to be identified includes a sensitization region;

the first sensitization area is perpendicular to the plane where the second sensitization area is located, the first sensitization area is parallel to the plane where the third sensitization area is located, and the first sensitization area is perpendicular to the plane where the fourth sensitization area is located.

In a possible implementation manner, the processing unit is further configured to perform coupling processing on two cores of the optical fiber to be identified; wherein the sensitized areas of the two cores are opposite.

In a third aspect, the present application provides a computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by an electronic device of the present application, cause the electronic device to perform the method of optical fiber bend angle identification as described in any one of the possible implementations of the first aspect and the first aspect.

In a fourth aspect, the present application provides an electronic device comprising: a processor and a memory; wherein the memory is configured to store one or more programs, the one or more programs comprising computer-executable instructions that, when executed by the electronic device, cause the electronic device to perform the method of identifying an angle of bend of an optical fiber as described in any one of the possible implementations of the first aspect and the first aspect.

In a fifth aspect, the application provides a computer program product comprising instructions which, when run on a computer, cause an electronic device of the application to perform the method of identifying a bend angle of an optical fiber as described in any of the possible implementations of the first aspect and the first aspect.

In a sixth aspect, the present application provides a chip system, the chip system being applied to an optical fiber bending angle recognition device; the system-on-chip includes one or more interface circuits, and one or more processors. The interface circuit and the processor are interconnected through a circuit; the interface circuit is configured to receive a signal from a memory of the optical fiber bend angle identification device and to send the signal to the processor, the signal including computer instructions stored in the memory. When the processor executes the computer instructions, the optical fiber bending angle identification device performs the optical fiber bending angle identification method according to the first aspect and any one of the possible designs thereof.

Drawings

FIG. 1 is a schematic diagram of polishing a side of an optical fiber to be identified according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a sensitization area of an optical fiber to be identified according to an embodiment of the present application;

FIG. 3 is a schematic diagram of an optical fiber to be identified after side polishing and coupling according to an embodiment of the present application;

FIG. 4 is a schematic diagram of an optical experiment platform according to an embodiment of the present application;

FIG. 5 is a schematic flow chart of a method for identifying an angle of bending an optical fiber according to an embodiment of the present application;

fig. 6 is a schematic flow chart of a convolutional neural network according to an embodiment of the present application;

FIG. 7 is a flowchart illustrating another method for identifying a bending angle of an optical fiber according to an embodiment of the present application;

FIG. 8 is a flowchart illustrating another method for identifying a bending angle of an optical fiber according to an embodiment of the present application;

FIG. 9 is a schematic structural diagram of an apparatus for identifying bending angle of an optical fiber according to an embodiment of the present application;

fig. 10 is a schematic structural diagram of another optical fiber bending angle recognition device according to an embodiment of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

The character "/" herein generally indicates that the associated object is an "or" relationship. For example, A/B may be understood as A or B.

The terms "first" and "second" in the description and in the claims of the application are used for distinguishing between different objects and not for describing a particular sequential order of objects. For example, the first edge service node and the second edge service node are used to distinguish between different edge service nodes, rather than to describe a characteristic order of the edge service nodes.

Furthermore, references to the terms "comprising" and "having" and any variations thereof in the description of the present application are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed but may optionally include other steps or elements not listed or inherent to such process, method, article, or apparatus.

In addition, in the embodiments of the present application, words such as "exemplary", or "such as" are used to mean serving as examples, illustrations, or descriptions. Any embodiment or design described herein as "exemplary" or "for example" should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary", or "such as" is intended to present concepts in a concrete fashion.

The technical terms related to the application are described below:

1. input/output power

After the plastic optical fiber is bent, the input and output power of the plastic optical fiber can be changed. The radiation loss of a plastic optical fiber after bending is generally characterized by the ratio of input power to output power.

In the embodiment of the application, the sensitization area is arranged on the optical fiber, and when the sensitization area is bent, the optical fiber cladding without the sensitization area is polished and removed by the side surface, so that the radiation loss is increased, and the ratio of input power to output power is changed. The specific input-output power ratio satisfies the following formula 1:

wherein Po represents input power, pi represents output power, S _C Represents the cross-sectional area of the sensitization region, S _O Represents the area of the cut surface of the unprocessed optical fiber, a represents the radius of the fiber core, R represents the radius of curvature, and theta _b Indicating the angle of bending of the optical fiber, θ _C Representing the critical angle obtained by snell's law.

2. Speckle pattern

In multimode fibers, the incident beam excites various radiation modes simultaneously, and the speckle actually observed from the output end is the result of superposition or interference of the speckle patterns of the modes. If an incoherent light source is used, the end of the optical fiber outputs uniformly distributed circular light spots, and if a coherent light source is used, the end of the optical fiber outputs interference speckle. Because the phase relation between the modes is affected by a plurality of factors, the coherent speckle can randomly shift, so that the light intensity in the local fiber core randomly fluctuates, and the source of the mode noise is formed.

In contrast, the numerical aperture of the plastic multimode fiber is larger, a large number of modes exist in the plastic multimode fiber, and a plurality of channels can simultaneously transmit information in parallel, so that high-resolution imaging is realized, and the image precision is improved. When the laser beam enters from the incident end, a seemingly random circular granular pattern, i.e. the aforementioned speckles, is formed at the multimode fiber output end due to the energy exchange and mutual interference between the transmission modes.

Each mode in the multimode fiber represents a pixel for carrying information about the pixel, but this contains the fiber state and the spatial information speckle pattern of the input signal, resulting in image distortion due to crosstalk caused by modal coupling and dispersion. The calculation formula of the number N of modes supported by the multimode optical fiber for transmission satisfies the following formula 2:

wherein V represents a normalized frequency, a _co Expressed as core radius, n _co And n _cl Denote the core and cladding refractive indices, NA denotes the numerical aperture, and N denotes the number of modes.

Many studies have shown that imaging or image transmission can be performed through multimode optical fibers, such as analog phase conjugation, digital iteration, or a method of constructing a complex matrix. When a laser beam with the wavelength lambda is incident on the plastic optical fiber, a plurality of modes excited equally interfere and generate speckles, and the calculated speckle intensity of the far-end output meets the following formula 3:

Wherein M represents the number of modes, a _m (x, y) andthe amplitude and phase of pattern m are represented, j represents the imaginary unit, exp represents an exponential function based on e.

Further, the intensity I (x, y) of the far-field speckle pattern of the optical fiber captured by the CCD camera satisfies the following equation 4:

wherein a is _m (x, y) and phi _m (x, y) represents the amplitude and phase of the mth mode, a, respectively _n (x, y) and phi _n (x, y) represents the amplitude and phase of pattern n.

And, when the optical fiber is bent, the deformation of the optical fiber causes energy transfer between modes, i.e., changes the distribution of speckles, specifically satisfying the following equation 5:

wherein delta _am And delta _φm Representing the amplitude and phase variation of the mth mode in the fiber, a _0m (x, y) and phi _0m (x, y). The amplitude and phase of the mth mode when the fiber is not bent are shown.

And, the speckle distribution intensity a (x, y) of the distal end of the optical fiber satisfies the following equation 6:

in summary, the speckle pattern intensity I (x, y) captured by the CCD satisfies the following equation 7:

where n represents the number of the nth pattern in the speckle pattern.

The principles involved in the speckle pattern acquisition experiment are described above.

3. Optical fiber coupling

According to the optical fiber mode coupling theory, when the original cladding and part of the fiber core structure of the optical fiber are damaged through side polishing sensitization treatment, the optical power can be coupled between two parallel optical fibers when the optical fiber is bent based on the light intensity modulation principle.

Based on the theory, the embodiment of the application provides a macrobend coupling system which is formed by carrying out side polishing sensitization on two fiber cores of a double-core plastic optical fiber and then arranging the two fiber cores in parallel, and realizes the coupling between the two optical fibers by utilizing the optical radiation effect caused by the bending of the optical fibers.

Because the guided mode in the fiber core is changed into a radiation mode to transmit, namely bending radiation loss occurs when the optical fiber is bent, based on the optical coupling theory, an effective coupling sensing system is formed if two fiber cores subjected to side polishing sensitization are placed together in parallel (two sensitization areas are opposite).

The technical terms related to the application are described above.

At present, when detecting the bending angle of a commercial optical fiber, a distributed sensing technology is generally adopted. Distributed sensing techniques typically use optical frequency domain reflectometry as well as optical time domain reflectometry, with resolution accuracy on the order of centimeters. However, the architecture of the sensing system using this technology is complex, the amount of hardware used is large, and therefore the cost is high, and the measurement range is relatively narrow.

Currently, sensors used in recognizing the bending angle of a plastic optical fiber (plastic optical fiber, POF) are generally classified into intensity-modulated sensors, wavelength-modulated sensors, interference-structure sensors, and the like. Specifically, the following describes the defects existing in the current scheme when the sensor is adopted:

(1) An intensity modulated sensor.

The intensity modulation type sensor is an optical fiber bending sensor based on intensity modulation. Although the sensor has the advantages of low cost, simple manufacturing process and the like, the sensor has a very obvious disadvantage of lacking multiplexing function. Specifically, intensity modulation based bending sensors cannot perform multi-point or multi-parameter measurements, which can reduce the flexibility and compactness of the sensing system if multiple optical fibers are used to perform the multi-point or multi-parameter measurements. In addition, most of the bending sensors in the prior art cannot distinguish between directions, or can distinguish between two opposite directions, and only a small portion can distinguish between multiple bending directions, which is insufficient for bending measurement in practical applications.

(2) Wavelength modulated sensors.

The wavelength modulation type sensor is an optical fiber bending sensor based on wavelength modulation, has high sensitivity, but simultaneously has the problem of cross sensitivity, and is easy to be influenced by practical application environment, namely, unexpected variables from the outside can cause interference on the measurement of bending angles.

Although wavelength-modulated and interferometric-based fiber-optic bending sensors have a high sensitivity, this also means that they suffer from cross-sensitivity, which is susceptible to practical application environments, i.e. undesired variables from the outside can interfere with the measurement of bending angles. In particular, optical fiber bending sensors based on wavelength modulation require high precision of the equipment and recovery of the output signal requires complex demodulation techniques and expensive spectroscopic measurement systems, such as spectroscopic analyzers. The high cost greatly restricts the wide application of the optical fiber bending sensor based on wavelength modulation.

(3) An interferometric structure type sensor.

The interferometric sensor has high sensitivity and cross sensitivity similar to that of the wavelength modulation type. That is, it is susceptible to the practical application environment, and undesired variables from the outside may interfere with the measurement of the bending angle.

And, interference structural sensor still has the complex system of supporting, and equipment is comparatively expensive, the measuring range is comparatively narrow problem.

The above description has been made with respect to the application of a large number of optical fiber bending sensors.

In addition, most of the bending sensors including the three sensors cannot distinguish directions or can distinguish only two opposite directions, and only a small part can distinguish a plurality of bending directions, which is insufficient for bending measurement in practical engineering use.

On the other hand, most of the bending sensors in the present stage detect the change of the optical power by a photodetector to calculate the bending angle, and this method does not consider the sensor error caused by the environmental factors, resulting in a larger error of the bending recognition result. On the other hand, in the present stage, the light source is fixed above most bending sensors, and the light source and the side polished surface are kept relatively static when the sensors work, so that the working range of bending the plastic optical fiber is smaller, and when the bending angle of the optical fiber is increased, the positions of the light source and the sensors are not easy to control, and the relative positions of each sensor and the light source are easy to be different, so that errors are increased.

On the other hand, the wavelength modulation type sensor and the interference structure type sensor have high requirements on the light source, and the common light source cannot realize high bending recognition effect.

Summarizing, it is known that the current stage has the problems of high cost, small measurable range, small measuring direction and large error when the bending angle of the optical fiber made of POF material is identified

In view of the above, the present application provides a method and apparatus for identifying an optical fiber bending angle, which can identify an optical fiber bending angle with low cost, large range, multiple directions and small error. The method for identifying the bending angle of the optical fiber provided by the application designs a step of polishing and grinding the side surface, so that the processed plastic optical fiber section to be identified becomes a distributed optical fiber sensor for measuring the bending angle of multiple points and multiple planes, and meanwhile, a speckle identification model is trained by utilizing a convolutional neural network, so that different bending angles and directions of plastic optical fibers can be identified in a classified manner based on a visible light source or a light-emitting diode (LED) lamp. The application does not need expensive optical instruments, and pretreats the plastic optical fiber, so that the optical fiber has extremely high sensitivity to a plurality of planes or directions, and can realize multi-directional bending angle identification. Because the application adopts the convolutional neural network to train the speckle image recognition model, the speckle image data of the plastic optical fiber on a charge-coupled device (CCD) camera is recognized according to the model, and the sensing error caused by the surrounding environment factors is overcome to a certain extent. Finally, because the optical fiber identified by the application is a double-core optical fiber, the optical fiber coupling method can be adopted to ensure that the light source and the side polished surface are kept relatively static when the sensor works, and the positions of the light source and the sensor basically cannot be changed when the bending angle of the optical fiber is increased. The method for identifying the bending angle of the optical fiber provided by the application is described in detail below with reference to the attached drawings:

Exemplary, as shown in fig. 1, fig. 1 is a schematic diagram of a side polishing process of an optical fiber to be identified according to the present application.

Wherein, part a of fig. 1 shows that a fiber core of an optical fiber to be identified is fixed on a plane, and the upper part of the fiber core is polished to form a sensitization area. The sensitization area of the polished optical fiber to be identified is shown as a part b in fig. 1, and the cross section of the sensitization area is similar to the shape of a capital letter D, so that the sensitivity of the optical fiber to be identified in bending deformation is improved.

In one possible implementation, the length of the sensitization zone is 20 millimeters and the polishing depth is one quarter of the core diameter of the fiber to be identified.

It should be understood that the length and polishing depth of the sensitization zone may be specifically set according to the actual application, and the exemplary description of the present application does not constitute limitation of the length and polishing depth of the sensitization zone.

Exemplary, as shown in fig. 2, fig. 2 is a schematic diagram of a sensitization area of an optical fiber to be identified provided in the present application.

It will be appreciated that the sensitivity angle of an unprocessed plastic fiber cannot distinguish between bending directions, and therefore requires a physical side polishing process.

Alternatively, as shown in fig. 2, the number of the sensitization regions is illustratively selected to be 4, and is respectively located in four equal areas of the core of a section of the optical fiber to be identified. Illustratively, a length of plastic optical fiber with a length of 60cm is selected from four equally divided regions of 15cm, 30cm, 45cm and 60cm from the input end to process two cores of a dual-core optical fiber.

It should be understood that the number of the sensitization regions may be specifically selected according to the actual application, and the exemplary description of the present application does not constitute a limitation on the selection of the number of sensitization regions.

And, as shown in part c of fig. 2, the first sensitization zone is perpendicular to the plane in which the second sensitization zone is located, the first sensitization zone is parallel to the plane in which the third sensitization zone is located, and the first sensitization zone is perpendicular to the plane in which the fourth sensitization zone is located. In one possible implementation, for a dual core fiber, as shown in part d of fig. 2, the present application couples the sensitized areas of the two cores relatively after side polishing each core of the plastic fiber.

It should be noted that, for more than two plastic optical fibers with fiber cores, the side polishing method provided by the application can be adopted to process the fiber cores and couple the fiber cores in pairs. That is, the side polishing process and the coupling process of the present application are also applicable to plastic optical fibers having two or more cores, for example, four-core optical fibers and eight-core optical fibers.

As shown in fig. 3, fig. 3 shows an optical fiber to be identified provided in the present application after being polished on a side surface and coupled.

In fig. 3, one core a of the dual-core plastic optical fiber is in a flat state, and the other core B is divided into four small segments for coupling. In the four small lengths of optical fiber, one side of each length is coupled over the fiber core A, and the other side is used for fixing the visible light source C.

It will be appreciated that the fiber to be identified shown in FIG. 3 includes four sensing zones, each corresponding to a separate sensor.

Exemplary, as shown in fig. 4, fig. 4 is an optical experiment platform 40 provided by the present application. The optical experiment platform comprises: the bottom panel 41, the optical clamp 42, the fixed end 43, the visible light source 44, the convex lens 45, the CCD camera 46 and the image acquisition device 47.

Wherein the bottom panel 41 is used for fixing other components of the optical experiment platform 40.

An optical clamp 42 for fixing the optical fiber to be identified. It will be appreciated that during the experiment, the optical clamp 42 can be moved up and down, left and right, such that the fiber to be identified is bent through an angle.

A fixed end 43, which is located above the optical fiber to be identified. That is, the light emitted from the visible light source 44 is incident on the optical fiber to be identified through the fixed end 43.

Alternatively, in the case where the pretreated optical fiber to be identified is embodied as shown in fig. 3, the number of the fixed ends 43 is 4.

It should be noted that, in order to obtain the speckle patterns of different sensitization areas of the optical fiber to be identified, the color of the visible light incident on each fixed end 43 is different. For example, the 4 fixed ends 43 respectively enter red, yellow, blue, and violet visible light.

A visible light source 44 for emitting visible light. It should be understood that the visible light sources 44 may be classified into various types, and the number of the visible light sources 44 is the same as the number of the fixed ends 43.

Alternatively, the visible light sources 44 may be implemented as LED lamps capable of emitting red, yellow, blue and green visible light, respectively located on the four fixed ends of the core B of the optical fiber to be identified as shown in fig. 3.

The convex lens 45 is used for focusing the light emitted by the optical fiber to be identified and making the light enter the CCD camera 46.

The CCD camera 46 is used for capturing the emergent light focused by the convex lens 45, forming a speckle pattern and outputting the speckle pattern to the image acquisition device 47.

The image acquisition device 47 is used for acquiring the speckle pattern output by the CCD camera 46 through built-in image acquisition software.

Alternatively, the image capturing device 47 may be implemented as a computer terminal, or other electronic device with image capturing functions.

The optical bench 40 provided by the present application was described above.

Illustratively, the type of the optical fiber to be identified in the application is multimode step-type POF, the core matrix is organic glass (methyl methacrylate, PMMA), the refractive index of the core is 1.49, the numerical aperture na=0.5, the outer diameter is 2×2.2mm, the inner diameter is 1mm, and the working temperature ranges from-55 degrees celsius to +70 degrees celsius.

It should be understood that in the process of collecting the speckle pattern, the bending angle and direction of the plastic optical fiber are changed by taking the sensitization area as the center, and the light intensity is correspondingly changed. Parallel light output by the tail end of the optical fiber to be identified is directly beaten on the CCD camera 46, and the speckle can fill the whole screen, so that a part of speckle information is lost, and if the speckle information is converged to a point through the convex lens 45 and then imaged to the CCD camera 46, the speckle with moderate size can be collected. Imaging information of the CCD camera 46 is collected by an image collecting device 47, and a speckle pattern is obtained by capturing a picture in real time by image collecting software.

It should be noted that, in order to ensure the accuracy and effectiveness of the experiment, the optical fiber to be identified is repeatedly bent before each screenshot, that is, the optical fiber to be identified bent to a certain angle is restored to a horizontal state, and then is bent to a previous angle again for collection. The purpose of doing so is to guarantee the absolute stability of experimental platform, reduces the output that causes the influence to the plastic optical fiber because of bending or shake of external factor.

In one possible implementation, the image acquisition software used in the experimental process adopts a TensorFlow deep learning open source framework, and a PyCharm compiler 3.7Python library is used for data processing by using a NVIDIA GeForce RTX 2070 graphic processing unit.

The principle of the coupling treatment of the optical fiber to be identified in the application is briefly described below:

Based on the theory, the theory proposes that after two fiber cores of the double-core plastic optical fiber are subjected to side polishing sensitization, the two fiber cores are placed in parallel to form a macrobend coupling system, and the coupling between the two optical fibers is realized by utilizing the optical radiation effect caused by the bending of the optical fiber.

Wherein in the bending region of the optical fiber, the light protruding toward the collection is refracted in the bending region, and a part of the power of the incident light is radiated from the single-mode optical fiber. The power of the radiated light satisfies the following equation 8:

P _o ＝P _i * T formula 8 wherein P _i Representing the power of the incident light, T represents the Fresnel transmission coefficient in the core, which satisfies the following equation 9:

wherein θ represents an incident angle, θ ₁ ＝[1-(n _eff /n _co ) ² ] ^1/2 Indicating the critical angle for total internal reflection.

As can be seen from the above equation 9, as the bending angle increases, the incident angle θ increases; at the same time, the core diameter alpha of the optical fiber can be known _co The larger the transmission coefficient T, the larger the radiation loss and the easier the light will be coupled into the plastic optical fiber.

And, according to the mode coupling theory, the coupling coefficient τ between two adjacent parallel fibers can satisfy the following equation 10:

wherein delta represents a Kroller function, U represents a normalized transverse phase parameter, W represents a transverse attenuation parameter, d represents a core wheelbase between two optical fibers, and alpha _co Represents the core diameter of the optical fiber, W (d/alpha) _co ) Represents the isolation degree between optical fibers, K ₁ 、K ₂ Representing Bessel function, V _norm Representing the normalized frequency. It can be seen that as d increases, the coupling coefficient τ decreases significantly. Thus, as the center-to-center distance becomes larger, the coupling capability between the fibers will be significantly reduced.

The principle of coupling the optical fibers to be identified is introduced.

The optical fiber bending angle recognition method according to the present application is mainly performed by an optical fiber bending angle recognition device, and the optical fiber bending angle recognition device may include a device having a preprocessing function, a device having a function of acquiring a speckle pattern obtained by the optical experiment platform, and a device for training a speckle pattern recognition model.

As shown in fig. 5, fig. 5 is a method for identifying a bending angle of an optical fiber according to the present application, which includes the following steps:

s501, the optical fiber bending angle recognition device preprocesses the optical fiber to be recognized and determines at least one sensitization area.

Optionally, the optical fiber bending angle recognition device performs pretreatment on the optical fiber to be recognized, which may be performing side polishing processing on the optical fiber to be recognized, so as to select at least one sensitization area. It should be noted that, the specific optical fiber bending angle identification device performs the side polishing processing on the optical fiber to be identified, see fig. 1 and fig. 2 and related description, and will not be repeated here.

Optionally, the type of optical fiber to be identified is a twin-core optical fiber. The optical fiber bending angle recognition device performs side polishing processing on each quarter area of two fiber cores of the optical fiber to be recognized, and each fiber core comprises four sensitization areas after being processed.

Optionally, the fiber bending angle recognition device is used for carrying out coupling treatment on the sensitization areas of the two fiber cores to obtain four sensitization areas, namely a first sensitization area, a second sensitization area, a third sensitization area and a fourth sensitization area.

Optionally, the first sensitization area is perpendicular to the plane in which the second sensitization area is located, the first sensitization area is parallel to the plane in which the third sensitization area is located, and the first sensitization area is perpendicular to the plane in which the fourth sensitization area is located.

Optionally, the length of the sensitization zone is 20 mm and the polishing depth is one quarter of the core diameter of the fiber to be identified.

S502, the optical fiber bending angle recognition device acquires speckle image data of the optical fiber to be recognized according to the optical experiment platform and at least one sensitization area.

Optionally, the optical experiment platform comprises: the device comprises a bottom panel, an optical clamp, a fixed end, a visible light source, a convex lens, a Charge Coupled Device (CCD) camera and an image acquisition device.

It should be noted that, the specific structure and experimental method of the optical experimental platform refer to fig. 4 and the related description, and are not repeated here.

Alternatively, the image acquisition device may acquire speckle image data of the optical fiber to be identified.

S503, the optical fiber bending angle recognition device inputs speckle image data of the optical fiber to be recognized into a trained speckle image recognition model, and determines a bending angle recognition result of the recognition optical fiber.

Alternatively, the speckle image recognition model is constructed from a convolutional neural network.

The examples are described below in connection with specific numerical examples:

in the experimental process, the bending angles of four sensitization areas are respectively acquired, the step length of the bending angles of the optical fibers is set to be 15 degrees, the speckle patterns of each sensitization area when bending by 0 degrees, 15 degrees, 30 degrees, 45 degrees, 60 degrees, 75 degrees and 90 degrees are sequentially acquired, the measuring range is from 0 degrees to 90 degrees, and 25 speckle patterns with different bending angles are acquired in total. 340 speckle patterns are acquired for each bending angle, and 8500 speckle patterns are acquired in total. The 8500 speckle images were divided into 3 groups of training, test and validation sets, which may be proportioned according to 0.6,0.2,0.2.

The size of the initial input picture of the speckle image recognition model is set to 64×64, and the size of the convolution kernel of the convolution neural network is 5×5. First, the convolution operation is performed, the number of convolution kernels of Conv1 and Conv2 is the same as 64, and then the output of the maximum pooling operation performed by the pooling layer is 32×32×64. Similarly, the outputs after Conv3, conv4, conv5, and Maxpooling2 having 128 convolution kernels are 16×16×128. The output feature map is then input to Conv5 for a total of 25 convolution kernels, followed by Batch Normalization (BN) processing. And finally, carrying out global average pooling on the 8×8×25 feature map after the Maxpooling3, adding all pixel values of the output feature map to obtain an average value, obtaining a 1×1×25 tensor, and outputting the tensor through softMax. The global average pooling replaces a full-connection layer, so that the number of parameters and the calculated amount are greatly reduced. The flow chart of the above experiment is shown in fig. 6.

In one possible implementation, the trained speckle image recognition model is determined according to the following steps: the optical fiber bending angle recognition device acquires a data set and performs data marking on the data set; then, the optical fiber bending angle recognition device adjusts parameters of the speckle image recognition model according to an optimizer and a loss function, wherein the optimizer is used for adjusting the learning rate of the speckle image recognition model; finally, under the condition that the loss function value meets the preset condition, the optical fiber bending angle recognition device determines that the speckle image recognition model training is completed. It should be noted that, the process of determining the trained speckle image recognition model by the specific optical fiber bending angle recognition device may be referred to the following S701-S703, and this implementation will not be repeated.

Based on the technical scheme, the application designs a side polishing step, so that the processed plastic optical fiber section to be identified becomes a distributed optical fiber sensor for measuring the multi-point multi-plane bending angle, and meanwhile, a speckle identification model is trained by utilizing a convolutional neural network, so that different bending angles and directions of plastic optical fibers can be identified in a classified manner based on a visible light source or a light-emitting diode (LED) lamp. The application does not need expensive optical instruments, and pretreats the plastic optical fiber, so that the optical fiber has extremely high sensitivity to a plurality of planes or directions, and can realize multi-directional bending angle identification.

As shown in fig. 7, in an exemplary embodiment of the method for identifying an angle of bending of an optical fiber according to the present application, a trained speckle image identification model is determined according to the following steps:

s701, the optical fiber bending angle recognition device acquires a data set, and performs data marking on the data set.

Optionally, the optical fiber bending angle recognition device performs preprocessing classification on the image used for training the model, and then performs data marking on the image to obtain a data set capable of being used for training the speckle image recognition model. In one possible implementation, the optical fiber bending angle identifying device sets the label of the 25 types of optical fiber bending angles in the form of a 25-bit list, wherein the first bit of the list is 1, the rest of the list is 0, the second bit of the list is 1, the rest of the list is 0, and similarly, the 25 th bit of the list is 0.

Optionally, the optical fiber bending angle recognition device can divide the data set into three groups of a training set, a testing set and a verification set according to a certain proportion. Each set of data sets is stored in a plurality of folders in sequence for subsequent reading. The training set is directly used for iterative training of the speckle image recognition model, the verification set is used for verifying the recognition accuracy of the speckle image recognition model in each iterative training, and the test set is used for testing whether the speckle image recognition model can be practically applied.

Alternatively, the training set, the test set, and the validation set may be in a 3:1:1 ratio.

S702, the optical fiber bending angle recognition device adjusts parameters of the speckle image recognition model according to the optimizer and the loss function.

The optimizer is used for adjusting the learning rate of the speckle image recognition model.

Alternatively, the optimizer may employ an adam optimizer. For example, a tf. Train. AdamOptimezer optimizer provided by TensorFlow may be specifically employed.

It can be understood that the optimizer adjusts the learning rate of each parameter to solve the minimum loss, so as to control the learning speed, so that the learning rate of each iteration has a definite range, the learning rate cannot become very large even if the gradient is large, and the parameter value is always kept in a relatively stable state.

It should be understood that the learning rate represents the speed of updating the weights of the neural network, when the learning rate is too high, the loss function iterates faster, oscillations are generated, the cost function skips the optimal solution, and when the learning rate is too low, the network updating speed may be slow, and the network takes a long time to converge. The method of setting different learning values can be generally adopted for training for multiple times to obtain a better value.

It should be noted that, in the speckle image recognition model of the present application, in the optimization algorithm for solving the neural network parameters, an optimization algorithm based on gradient descent is mainly used, and specifically, a random gradient descent method (stochastic gradient descent, SGD) is adopted.

One possible example of the loss function involved in this embodiment is described below:

alternatively, the loss function satisfies the following equation 11:

wherein m represents the number of training samples, n represents the number of features, (θ) ₁ ,θ ₂ ,...θ _n )、(x ₁ ,x ₂ ,...x _n ) Representing each sampleCharacteristic value of the root, h _θ (x) Represents a regression equation, θ represents a weight, y _i Representing the output samples.

It should be appreciated that to minimize the value of the loss function J (θ), an optimal solution for θ is required. The central idea of the gradient descent algorithm is: firstly, defining an initial value for theta randomly, and then, selecting a direction with the largest change of J (theta) to continuously update the iterative value of theta so that the value of J (theta) becomes smaller until the loss function J (theta) is approximately equal to the minimum value, wherein the following formula 12 is specifically satisfied:

Wherein, alpha represents step length in the formula, which is used for controlling the change amplitude of theta when updating towards the direction of decreasing the loss function J (theta), and theta represents partial derivative. The bias of J (θ) to θ represents the direction in which J (θ) varies the most. And because the solution is the minimum, the gradient direction is the opposite direction of the partial derivative. The iterative formula for θ satisfies the following formula 13:

in the method, in the process of the invention,representative of sample features, where (j=0, 1,., m), y _i Representing the output samples. When the SGD performs network optimization, instead of scanning all training data sets, a data sample is randomly selected in the direction of the fastest calculation drop to calculate, and the data sample tends to be very small in an oscillating mode, so that the iteration speed is increased, but unstable parameter updating can be caused.

After the parameters are set, the collected speckle data atlas and the label can be added into a convolutional neural network written in a TensorFlow framework to be unfolded and trained, and a speckle image recognition model obtained after training is stored.

One possible example of the loss function involved in the present embodiment is described above.

It can be appreciated that in this embodiment, when the speckle image recognition model is trained for the first time, parameters to be trained of the convolutional neural network are initialized, the training iteration number is set to be 200, and the value of Batch training sample Batch size is set to be 500. The specific value is set reasonably according to the complexity of the convolutional neural network and the hardware condition of the specific computer.

In one possible implementation manner, the optical fiber bending angle recognition device further calculates a model error and a weight modification quantity through a back propagation algorithm, updates a weight value omega, wherein omega is a weight parameter, takes a value as a constant, and takes a value according to the actual condition of optimization of a specific program.

Therefore, the optical fiber bending angle recognition device optimizes the speckle image recognition model based on the weight parameter omega, and searches the optimal parameter so that the loss function value of the speckle image recognition model is minimum.

S703, the optical fiber bending angle recognition device determines that the speckle image recognition model training is completed under the condition that the loss function value meets the preset condition.

Optionally, the preset condition is that the loss function reaches convergence, or that the loss function reaches a minimum value in a preset number of iterations.

Optionally, the preset condition may be that the recognition accuracy of the speckle image recognition model in the training process meets the preset accuracy.

Based on the technical scheme, when the speckle image data on the CCD camera is processed, the speckle image recognition model obtained through convolutional neural network training is adopted, so that sensing errors caused by surrounding environment factors can be overcome, and meanwhile, the collected speckle images are classified through the speckle image recognition model, a complex detection system is not needed, and the scheme cost is saved.

Exemplary, referring to fig. 5, and referring to fig. 8, the method for identifying a bending angle of an optical fiber according to the present application further includes the following steps:

s801, the optical fiber bending angle recognition device performs coupling treatment on two fiber cores of the optical fiber to be recognized.

Optionally, the sensitized areas of the two cores are coupled relatively.

It should be noted that, the principle of the coupling processing of the optical fiber to be identified in this embodiment may be referred to the description of the foregoing formula 8-formula 10, and this embodiment is not repeated here.

It should be understood that, in this embodiment, the optical fiber coupling method is adopted to keep the light source and the side polished surface relatively static when the pretreated optical fiber to be identified works, and when the bending angle of the optical fiber increases, the positions of the light source and the sensor are not changed basically.

Based on the technical scheme, the optical fiber coupling method is adopted to enable the light source and the side polished surface to keep relatively static when the sensor works, and when the bending angle of the optical fiber is increased, the positions of the light source and the sensor are basically unchanged.

The embodiment of the application can divide the functional modules or functional units of the optical fiber bending angle identification device according to the method example, for example, each functional module or functional unit can be divided corresponding to each function, and two or more functions can be integrated in one processing module. The integrated modules may be implemented in hardware, or in software functional modules or functional units. The division of the modules or units in the embodiment of the present application is schematic, which is merely a logic function division, and other division manners may be implemented in practice.

Illustratively, as shown in fig. 9, a schematic diagram of a possible structure of an optical fiber bending angle identifying device according to an embodiment of the present application is shown. The optical fiber bending angle recognition apparatus 900 includes: processing section 901 and acquiring section 902.

The processing unit 901 is used for preprocessing the optical fiber to be identified and determining at least one sensitization area;

the processing unit 901 is further configured to obtain speckle image data of an optical fiber to be identified according to the optical experiment platform and the at least one sensitization area;

the processing unit 901 is further configured to input the speckle image data of the optical fiber to be identified into a trained speckle image identification model, and determine a bending angle identification result of the identified optical fiber.

Optionally, the acquiring unit 902 is configured to acquire a data set, and perform data marking on the data set.

Optionally, the processing unit 901 is further configured to adjust parameters of the speckle image recognition model according to the optimizer and the loss function; the optimizer is used for adjusting the learning rate of the speckle image recognition model;

optionally, the processing unit 901 is further configured to determine that the training of the speckle image identification model is completed when the loss function value meets a preset condition.

Optionally, the processing unit 901 is further configured to perform coupling processing on two cores of the optical fiber to be identified; wherein the sensitized areas of the two cores are opposite.

Alternatively, the optical fiber bending angle identification device 900 may further include a storage unit (shown in a dashed line box in fig. 9) storing a program or instructions that, when executed by the processing unit 901 and the acquiring unit 902, enable the optical fiber bending angle identification device to perform the optical fiber bending angle identification method described in the above-described method embodiment.

In addition, the technical effects of the optical fiber bending angle identifying device shown in fig. 9 may refer to the technical effects of the optical fiber bending angle identifying method described in the foregoing embodiments, and are not repeated herein.

Illustratively, fig. 10 is a schematic view of still another possible structure for identifying the bending angle of the optical fiber according to the above embodiment. As shown in fig. 10, the optical fiber bending angle identification 1000 includes: a processor 1002.

The processor 1002 is configured to control and manage the operation of identifying the bending angle of the optical fiber, for example, perform the steps performed by the processing unit 901 and the obtaining unit 902 in the optical fiber bending angle identifying device 900 and/or perform other processes of the technical solutions described herein.

The processor 1002 may be implemented or realized with the various illustrative logical blocks, modules, and circuits described in connection with the present disclosure. The processor may be a central processing unit, a general purpose processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Which may implement or perform the various exemplary logic blocks, modules and circuits described in connection with this disclosure. The processor may also be a combination that performs the function of a computation, e.g., a combination comprising one or more microprocessors, a combination of a DSP and a microprocessor, etc.

Optionally, the fiber bend angle identification 1000 may further include a communication interface 1003, a memory 1001, and a bus 1004. Wherein the communication interface 1003 is used to support communication of the fiber bend angle identification 1000 with other network entities. Memory 1001 is used to store program codes and data for this fiber bend angle identification.

Wherein the memory 1001 may be a memory in fiber bend angle identification, which may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk or solid state disk; the memory may also comprise a combination of the above types of memories.

Bus 1004 may be an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus or the like. The bus 1004 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in fig. 10, but not only one bus or one type of bus.

From the foregoing description of the embodiments, it will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of functional modules is illustrated, and in practical application, the above-described functional allocation may be implemented by different functional modules according to needs, i.e. the internal structure of the apparatus is divided into different functional modules to implement all or part of the functions described above. The specific working processes of the above-described systems, devices and modules may refer to the corresponding processes in the foregoing method embodiments, which are not described herein.

An embodiment of the present application provides a computer program product containing instructions, which when run on an electronic device of the present application, cause the computer to perform the offload policy determination method described in the above embodiment of the method.

The embodiment of the application also provides a computer readable storage medium, wherein the computer readable storage medium stores instructions, and when the computer executes the instructions, the electronic equipment executes each step of the optical fiber bending angle identification execution in the method flow shown in the embodiment of the method.

The computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: electrical connections having one or more wires, portable computer diskette, hard disk. Random access Memory (Random Access Memory, RAM), read-Only Memory (ROM), erasable programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), registers, hard disk, optical fiber, portable compact disc Read-Only Memory (CD-ROM), an optical storage device, a magnetic storage device, or any other form of computer-readable storage medium suitable for use by a person or persons of skill in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an application specific integrated circuit (Application Specific Integrated Circuit, ASIC). In embodiments of the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The foregoing is merely illustrative of specific embodiments of the present application, and the scope of the present application is not limited thereto, but any changes or substitutions within the technical scope of the present application should be covered by the scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for identifying an angle of bend of an optical fiber, the method comprising:

preprocessing the optical fiber to be identified, and determining at least one sensitization area;

according to the optical experiment platform and the at least one sensitization area, speckle image data of the optical fiber to be identified are obtained;

and inputting the speckle image data of the optical fiber to be identified into a trained speckle image identification model, and determining the bending angle identification result of the identification optical fiber.

2. The method of claim 1, wherein the optical bench comprises: the device comprises a bottom panel, an optical clamp, a fixed end, a visible light source, a convex lens, a Charge Coupled Device (CCD) camera and an image acquisition device;

the bottom panel is used for fixing the optical clamp, the fixed end, the visible light source, the convex lens, the CCD camera and the image acquisition device;

The optical clamp is used for fixing the optical fiber to be identified;

the fixed end is positioned on the optical fiber to be identified, and the fixed end and the convex lens are respectively positioned at two sides of the optical clamp; the visible light emitted by the visible light source is emitted into the optical fiber to be identified from the fixed end;

the CCD camera is used for capturing emergent light emitted from the convex lens;

the image acquisition device is connected with the CCD camera and used for acquiring speckle image data of the optical fiber to be identified.

3. The method of claim 2, wherein the speckle image recognition model is constructed from a convolutional neural network, and wherein the trained speckle image recognition model is determined from:

acquiring a data set, and carrying out data marking on the data set;

according to the optimizer and the loss function, adjusting parameters of the speckle image recognition model; the optimizer is used for adjusting the learning rate of the speckle image recognition model;

and under the condition that the loss function value meets the preset condition, determining that the speckle image recognition model training is completed.

4. A method according to claim 3, wherein the pre-processing comprises: and carrying out side polishing processing on the optical fiber to be identified, and selecting at least one sensitization area.

5. The method of claim 4, wherein said at least one plenum region comprises a first plenum region, a second plenum region, a third plenum region, a fourth plenum region, and four of said plenum regions, each quarter of said fiber to be identified comprising a plenum region;

6. The method of claim 5, wherein each of the sensitized areas has a length of 20 millimeters and a polishing depth of one quarter of a core diameter of the optical fiber to be identified.

7. The method of claim 6, wherein the type of fiber to be identified is a dual core plastic fiber; each fiber core of the optical fiber to be identified is polished and processed through the side surface, and each fiber core comprises four sensitization areas.

8. The method according to any one of claims 1-7, wherein prior to said acquiring speckle image data of said fiber to be identified from said at least one sensitization zone and optical bench, the method further comprises:

Coupling the two fiber cores of the optical fiber to be identified; wherein the sensitized areas of the two cores are opposite.

9. An optical fiber bending angle recognition apparatus, characterized in that the optical fiber bending angle recognition apparatus includes: a processing unit;

the processing unit is used for preprocessing the optical fiber to be identified and determining at least one sensitization area;

the processing unit is further used for acquiring speckle image data of the optical fiber to be identified according to the optical experiment platform and the at least one sensitization area;

the processing unit is further used for inputting the speckle image data of the optical fiber to be identified into a trained speckle image identification model, and determining the bending angle identification result of the identification optical fiber.

10. The fiber bend angle identification device of claim 9, wherein the optical experiment platform comprises: the device comprises a bottom panel, an optical clamp, a fixed end, a visible light source, a convex lens, a Charge Coupled Device (CCD) camera and an image acquisition device;

The optical clamp is used for fixing the optical fiber to be identified;

11. The optical fiber bending angle identification device according to claim 10, further comprising: an acquisition unit;

the acquisition unit is used for acquiring a data set and marking the data set;

the processing unit is also used for adjusting parameters of the speckle image recognition model according to the optimizer and the loss function; the optimizer is used for adjusting the learning rate of the speckle image recognition model;

and the processing unit is also used for determining that the speckle image recognition model training is completed under the condition that the loss function value meets the preset condition.

12. The fiber bend angle identification device of claim 11, wherein the preprocessing comprises: and carrying out side polishing processing on the optical fiber to be identified, and selecting at least one sensitization area.

13. The fiber bend angle identification device of claim 12, wherein said at least one plenum zone comprises a first plenum zone, a second plenum zone, a third plenum zone, a fourth plenum zone, and four of said plenum zones, each quarter of said fiber to be identified comprising a plenum zone;

14. The fiber bend angle identification device of claim 13, wherein each of said sensitized areas has a length of 20 millimeters and a polishing depth of one quarter of a core diameter of said fiber to be identified.

15. The fiber bend angle identification device of claim 14, wherein the type of fiber to be identified is a two-core plastic fiber; each fiber core of the optical fiber to be identified is polished and processed through the side surface, and each fiber core comprises four sensitization areas.

16. The optical fiber bending angle identification device according to any one of claims 9 to 15, wherein,

The processing unit is also used for carrying out coupling processing on the two fiber cores of the optical fiber to be identified; wherein the sensitized areas of the two cores are opposite.

17. An electronic device, comprising: a processor and a memory; wherein the memory is configured to store computer-executable instructions that, when the electronic device is in operation, cause the electronic device to perform the method of identifying an angle of bend of an optical fiber as claimed in any one of claims 1 to 8.

18. A computer readable storage medium comprising instructions that, when executed by an electronic device, enable the electronic device to perform the method of identifying an angle of bend of an optical fiber as claimed in any one of claims 1-8.